Analysis of Speech Separation Performance Degradation on Emotional Speech Mixtures
2023 ASIA PACIFIC SIGNAL AND INFORMATION PROCESSING ASSOCIATION ANNUAL SUMMIT AND CONFERENCE, APSIPA ASC(2023)
摘要
Despite recent strides made in Speech Separation, most models are trained on datasets with neutral emotions. Emotional speech has been known to degrade performance of models in a variety of speech tasks, which reduces the effectiveness of these models when deployed in real-world scenarios. In this paper we perform analysis to differentiate the performance degradation arising from the emotions in speech from the impact of out-ofdomain inference. This is measured using a carefully designed test dataset, Emo2Mix, consisting of balanced data across all emotional combinations. We show that even models with strong out-of-domain performance such as Sepformer can still suffer significant degradation of up to 5.1 dB SI-SDRi on mixtures with strong emotions. This demonstrates the importance of accounting for emotions in real-world speech separation applications.
更多查看译文
关键词
speech separation,transformer,deep learning,emotional speech,emotion classification
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要